summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Matthew's patch: don't add requeue'd messages to unconfirmedAlexandru Scvortov2011-01-141-2/+6
|
* merge bug23678 into default (conditional on vq:needs_idle_timeout wrt ↵Alexandru Scvortov2011-01-131-29/+23
|\ | | | | | | confirms is too lax)
| * merge bug23684 into defaultMatthias Radestock2011-01-136-208/+210
| |\
| * | remove redundant clauseMatthias Radestock2011-01-131-2/+0
| | | | | | | | | | | | | | | confirm/3 is now only ever called once confirm mode has been enabled, and since we can't turn that mode off there is no point in checking.
| * | generalise optimisationMatthias Radestock2011-01-131-6/+6
| | | | | | | | | | | | | | | there is never any point in trying to send a multi-ack when there is just one message to confirm
| * | bug fix: don't send confirms when MsgSeqNo == undefinedMatthias Radestock2011-01-131-7/+13
| | | | | | | | | | | | | | | | | | | | | also optimise unrouted message handling further - there is no point in going through all the logic of determining whether we can send a multi-ack since we know the MsgSeqNo will be the highest unconfirmed and it's just a single message we are confirming.
| * | cosmeticMatthias Radestock2011-01-131-13/+10
| | |
| * | inline a functionAlexandru Scvortov2011-01-131-11/+3
| | |
| * | optimise confirms for unroutable messages; don't do any unnecessary set/dict ↵Alexandru Scvortov2011-01-131-12/+13
| | | | | | | | | | | | operations
* | | merge from defaultAlexandru Scvortov2011-01-139-261/+275
|\ \ \ | | |/ | |/|
| * | Backed out changeset b6d2a6e12e77Matthias Radestock2011-01-131-0/+2
| | | | | | | | | | | | Turns out this change is only valid on the bug23684 branch
| * | remove redundant clauseMatthias Radestock2011-01-131-2/+0
| | | | | | | | | | | | | | | confirm/3 is only ever called once confirm mode has been enabled, and since we can't turn that mode off there is no point in checking
| * | tiny refactor: use dict:fetch instead of dict:findMatthias Radestock2011-01-131-2/+2
| | |
| * | Merging bug23643 into defaultMatthew Sackman2011-01-121-21/+31
| |\ \ | | |/
| | * Merging bug23668 into defaultMatthew Sackman2011-01-121-21/+31
| | |\
| | | * Refactoring and ensure that we don't issue confirms for messages that have ↵Matthew Sackman2011-01-121-24/+23
| | | | | | | | | | | | | | | | been masked from going to disk
| | | * msg_store confirms messages that are written to disk, removed and then revivedAlexandru Scvortov2011-01-121-11/+22
| | | |
| * | | Eliminate unnecessary put/2, and cosmeticMatthew Sackman2011-01-121-5/+4
| | | |
| * | | typoMatthew Sackman2011-01-121-1/+1
| | | |
| * | | merge default into bug23643Matthias Radestock2011-01-1229-759/+1269
| |\ \ \ | | |/ /
| | * | merge headsMatthias Radestock2011-01-121-16/+16
| | |\ \
| | | * | cosmetic: fix indentationMatthias Radestock2011-01-121-16/+16
| | | | |
| | * | | Mandatory delivery only needs to know that it got to the queue. Not the ↵Matthew Sackman2011-01-121-4/+5
| | |/ / | | | | | | | | | | | | result of delivery
| | * | Prioritise both sync and async versions the sameMatthew Sackman2011-01-121-12/+13
| | |/
| * | refactor: extract function to handle frame and pass to channelMatthias Radestock2011-01-061-23/+24
| | |
| * | cosmeticMatthias Radestock2011-01-061-7/+7
| | |
| * | cosmeticMatthias Radestock2011-01-031-2/+2
| | |
| * | better error handlingMatthias Radestock2011-01-031-5/+8
| | |
| * | add specs and move frame analysis from reader to command assemblerMatthias Radestock2011-01-032-22/+57
| | | | | | | | | | | | | | | The latter avoids a mutual dependency between the reader and command assembler.
| * | renamingMatthias Radestock2011-01-033-24/+24
| | |
| * | replace rabbit_framing_channel with tiny state machineMatthias Radestock2011-01-014-103/+66
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ...which we store in the reader's process dict This simplifies the interactions between the channel and the reader. Note that the reader now monitors the channel processes rather than their sups. This makes more sense since it interacts with the former but never the latter.
| * | refactoring in preperation for eliminating the framing_channel processMatthias Radestock2011-01-011-68/+69
| | |
* | | Logic.Matthew Sackman2011-01-121-4/+1
| | |
* | | Make conditional more preciseMatthew Sackman2011-01-121-9/+26
| |/ |/|
* | fix a bugMatthias Radestock2011-01-111-4/+7
| |
* | cosmeticMatthias Radestock2011-01-111-4/+3
| |
* | cosmetic and tiny refactorMatthias Radestock2011-01-111-5/+5
| |
* | Optimise the heartbeat to death. Honk. And some cosmetics tooMatthew Sackman2011-01-112-11/+3
| |
* | merge bug23329 into defaultMatthias Radestock2011-01-111-1/+1
|\ \
| * \ Merging bug23616 into defaultMatthew Sackman2011-01-111-1/+1
| |\ \
| | * | start confirm sequence numbering at 1 instead of 0Matthias Radestock2011-01-111-1/+1
| |/ / | | | | | | | | | in order to avoid conflict with special meaning of deliver_tag=0 in basic.ack
* | | TyposMatthew Sackman2011-01-111-12/+12
| | |
* | | Updated documentationMatthew Sackman2011-01-111-5/+60
| | |
* | | refactor: extract helper functionMatthias Radestock2011-01-111-15/+12
| | |
* | | fix typoMatthias Radestock2011-01-111-1/+1
| | |
* | | cosmeticMatthias Radestock2011-01-111-6/+5
| | |
* | | fixMatthias Radestock2011-01-111-4/+4
| | |
* | | merge from defaultMatthias Radestock2011-01-118-173/+207
|\ \ \ | |/ /
| * | merge bug23593 into defaultMatthias Radestock2011-01-102-129/+93
| |\ \
| | * | small simplifying refactorMatthias Radestock2011-01-101-16/+15
| | | |
View Lorry Controller Status